Cellular Separation via Dielectrophoretic Field-Flow-Fractionation
Keith Dvorkin
ABSTRACT
I present a brief review of a cellular separation technique known as dielectrophoretic field-flowfractionation. The technique relies upon suspending different cells at different heights within a
laminar flow profile. The differentially suspended cells, thus, travel at different speeds and elute
from the flow chamber at different times resulting in successful separation. An alternate
fabrication procedure than that commonly described in the literature is then proposed utilizing a
micromolded polydimethylsiloxane elastomer cast over a thick photoresist. The proposed
procedure decreases the number of steps in the fabrication meanwhile providing more precise
control over the dimensions of the resultant device.
INTRODUCTION
The isolation and separation of distinct cell
populations from a suspended mixture, a process
known as fractionation, presents a challenging and
routinely encountered problem in cell biology,
molecular genetics, clinical diagnostics and
therapeutics.1
The most commonly used
technique for separation of cell populations is
centrifugation, which relies on differences in cell
density to cause layered sedimentation of different
cell populations. This technique has served rather
well for procedures designed to deplete or enrich
certain cell populations in many settings, but as
this and other successful fractionation techniques
have reached maturity, progress in the separation
resolution, cell purity, sample size, device cost,
and portability have been relatively hard to come
by.1 Dielectrophoretic field-flow-fractionation or
(DEP-FFF) is a recently developed (1997) 2 cell
separation technique which relies on the intrinsic
differences in cellular dielectric constants to
suspend cells at different heights in a fluid
experiencing laminar flow. Two cell populations
suspended in a thin chamber, with their inherent
difference in dielectric constant and corresponding
difference in equilibrium height, would therefore,
experience different velocities in the familiar
parabolic velocity profile of laminar flow, and
would thus exhibit different elution times from this
chamber. The basic principle of DEP-FFF is
illustrated in fig. 1. This technique is rather simple
to implement, may be done so under sterile
conditions, and also provides excellent separation
resolution. Since it relies on intrinsic differences in
cellular populations rather than externally
generated cues such as in fluorescence- or
magnetic-activated cell sorting, it offers additional
comfort to the user by ensuring that the cells
exposed in the DEP-FFF chamber are in their
normal metabolic state after passing through the
separation system.3 The DEP-FFF system thus
has potential applications in the separation of
cancer cells from bone marrow or mobilized blood
in the treatment of advanced cancers in which
autologous hematopoietic cell transplantation is
required.1 The DEP-FFF system also finds use in
detecting cancer cells circulating in the peripheral
blood, a diagnostic tool in the early detection of
cancer.
Finally, DEP-FFF provides a cell
separation technique that is accessible from the
microscale, and one that me be serially
implemented in integrated microfluidic analysis
devices such as the so called micro-total-analysissystems or (μ-TAS).1
Fig. 1.1 Schematic depiction of the operational principle of
DEP-FFF.
Particles (or cells) with distinctive dielectric
constants will suspend themselves at different equilibrium
heights within the parabolic velocity profile of the advancing
liquid, and will thus travel at different velocities. This differential
in the velocity of traveling cells results in different elution times
for different cell populations resulting in separation of the
populations.
THEORETICAL
BACKGROUND
(appropriately
watered down)
Following Gayscoyne et al., fig. 1 shows
the setup of a typical DEP-FFF chamber, and the
forces acting on particles in solution. FDEPz is the
dielectrophoretic levitation force provided from the
interdigitated array of electrodes on the bottom
surface of the chamber. Under AC electrical
excitation, particles with a different dielectric
coefficient than their suspending medium will
experience a force due to the polarization of the
particle.
That force may be either positive
(attractive) or negative (repulsive) and depends on
a number of variables including the frequency of
the excitation and the dielectric constants of the
medium and particle. Additionally, this DEP force
decreases with the height of the particle over the
electrode array. Thus, when the frequency is
adjusted such that the DEP force on a particular
particle or set of particles is negative (repulsive)
that particle will suspend itself over the electrode
array at the exact height where the repulsive DEP
force just balances the sedimentation force, Fsed,
in the opposite direction. Since different cell
populations
exhibit
different
characteristic
dielectric constants, the equilibrium height, heq, at
which the sedimentation and DEP forces balance
will be unique to a particular cell type. When
subjected to a parabolic flow velocity profile, this
height variation may be exploited to physically
separate the cell populations. Equations exist
which allow one to predict the equilibrium height of
suspended particles, and one may therefore,
predict the elution times for a specified chamber
geometry and fluid flow rate however, for our
purposes a general understanding of the
underlying physics will suffice.
Different cells exhibit characteristic
differences in dielectric constant due mainly to
differences manifest in the morphology of the cell
membrane. The membrane capacitance, directly
related to the polarizability of the cell as a whole,
is found to increase with area of the cell
membrane.4
As
different
cells
exhibit
characteristic membrane textures, ie. some cells
have smooth surfaces while others exhibit a
characteristic degree of folding and ruffles, the
membrane capacitance and polarizability vary
accordingly. In addition differences in membrane
composition inherent among different cell types in
also thought to influence the cell dielectric
constant further discriminating among cell types.4
EXAMPLE:
Separation
of
normal
Tlymphocytes from human breast cancer cells
at a demanding ratio
To further demonstrate the principle of
operation as well as to provide some operational
data and detail from a working DEP-FFF system
we will look at the performance of the device in a
specific cellular separation problem, that of
separating breast cancer cells from T-cells. As
previously mentioned, this separation is clinically
relevant as a screening tool in the early detection
of cancer cells present in the peripheral
bloodstream.
Fig. 21 Cell count vs. elution time for A) T-lymphocytes and B)
MDA-436 human breast cancer cells as a function of applied
frequency. Both cell populations demonstrated a narrow
elution peak at 5kHz with similar elution times. However, as
the frequency was increased above 10kHz the MDA-436
elution peak rapidly broadened while that of the T-lymphocytes
remained relatively narrow.
Cell fractograms (cell counts as determined by
flow cytometry) as a function of applied frequency
and elution time are shown in figure 2 for both Tcells and breast cancer cells (MDA-436). One
immediately notices that both cell populations
exhibit a sharp peak at 5kHz centered about 5 min
of elution time. However, as the applied frequency
increases over 10kHz the MDA-436 peak rapidly
broadens while the T-cell counts remain relatively
narrow with respect to elution time.
This
broadening results mainly from the near zero or
positive DEP forces experienced by the MDA-436
cells at frequencies above ~20kHz. These small
forces levitate the cells at very small equilibrium
heights and thus low velocities, or tend to trap the
breast cancer cells at the electrodes, while
allowing the T-lymphocytes which experience
negative DEP forces to be eluted from the
chamber. After the complete elution of the T-cells,
the remaining breast cancer cells may then be
released from the chamber by decreasing the
applied frequency. Cells mixed at an initial ratio of
2:3 were separated within 11 minutes with purity
above 92% at 30kHz.1 Additional cell mixtures
have also been successfully separated using
DEP-FFF with even higher purity.1-3, 5
FABRICATION PROCEDURE
Although the device depicted in fig. 1 is by
no means complicated its fabrication nevertheless
involves several time consuming steps that one
may eliminate with a few simple modifications to
the procedure. The DEP-FFF system shown in
figure 1 consists of a gold microelectrode array on
a glass substrate fabricated by standard
photolithographic patterning and liftoff techniques.
The electrodes have 50 μm spacing and width and
are pattered onto a 50X50 mm glass substrate. A
Teflon spacer 420 μm thick, cut in the center to
provide the separation channel (25X388 mm), is
then sandwiched between the electrode array and
an additional top glass plate. The two glass plates
have between them three 1.6 mm holes drilled
such that fluid and cells may be infused and
extracted through attached tubing, and the whole
construction must be tightly clamped together to
avoid fluid leakage.
My proposed alterations to the fabrication
procedure would change this three-layer structure
to a two-layer structure in which no holes need to
be drilled, and forceful clamping is unnecessary.
In addition, one may define the geometry of the
separation channel with lithographic precision.
The alterations hinge on two key materials which
exhibit unique properties, properties which have
made then invaluable in much current microfluidic
research. These materials are the ultra-thick
negative SU-8 photoresists (Microchem, MA) and
polydimethylsiloxane (PDMS).
PDMS is a
remarkable material available from Dow Corning
as a two part kit: silicone oil plus a curing agent.
When the curing agent is mixed with the silicone
oil, it serves to crosslink the long molecular chains
of the oil into an insoluble network, forming an
elastomer in the process. The resulting rubbery
material, similar in feel to a superballI, is optically
transparent down to 300 nm, and when used in
micromolding applications such as those in
microcontact-printing (G. Whitesides), PDMS has
been shown to accurately replicate features down
to the nanometer scale. The inherent elasticity of
PDMS allows the material to easily form conformal
contacts that are watertight at most pressures
encountered in microfluidic systems, and also
facilitates release from micromolds without
damage. SU-8 is an equally remarkable material
which has found extensive use in the field of
MEMS and microfluidics for its ability to easily
form high aspect ratio microstructures by doing
nothing
more
than
conventional
contact
photolithography.
SU-8, available in different
thickness formulations, is merely an extremely
high viscosity negative photoresist. It may be spin
coated onto substrates in thicknesses exceeding
200 μm in a single spin coating step. When
exposed to i-line (365nm) UV light the resist
rapidly crosslinks to form a chemically resistant
and thermally stable epoxy, that develops in
solution to form structures with nearly vertical
sidewalls.5
Fig. 3. Schematic depiction of the proposed fabrication
process for a simple DEP-FFF system. a) Bare Si wafer. b)
Thick photoresist patterned atop Si. c) PDMS molded and
cured over SU-8 on Si mold. Right: Section of the final device
including syringe ports for the introduction and extraction on
cell suspensions.
As one may envision from the unique
properties of these two materials my proposed
fabrication procedure involves casting PDMS
prepolymer mixed with curing agent over a
microfabricated mold consisting of
SU-8
photoresist on silicon. After curing, the PDMS
elastomer may then be peeled off of the mold and
placed on top of an interdigitated electrode array
on glass similar to that shown in figure 1. Since
the PDMS forms watertight conformal contacts
only gentle pressure will be required to keep the
PDMS in place. In addition, there will be no need
to drill infusion and extraction ports as PDMS is
readily penetrated by syringes and forms selfsealing punctures. SU-8 micromolding is ideal for
forming intricate fluidic patterns the kind which are
likely necessary in microsystems such as the μTAS. The proposed fabrication procedure and a
section of the final device is shown schematically
in figure 3.
CONCLUSION
DEP-FFF is a versatile technique for
cellular separation. By separating cells based on
their intrinsic differences it ensures that the
separated cells are not affected by the
fractionation procedure, and are thus available
after separation for further study. In addition the
separation technique avoids the use of bulky
hardware such as centrifuges and flow cytometers
which have been found difficult to implement in the
microscale.
DEP-FFF is, therefore, an ideal
technique for use in microfluidic applications such
as the μ-TAS in which cell separation is a critical
step in device processing.
REFERENCES
1. X. Wang, Anal. Chem., 72(4) Feb. 2000, 832-9
2. Y. Huang, Biophys. J., 73 Aug. 1997, 1118-29
3. J. Yang, Biophys. J., 78 May. 2000, 2680-9
4. J. Yang, Biophys. J., 76 Jan. 1999, 3307-14
5. J. Yang, Anal. Chem., 71(5) Mar. 1999, 911-8
6. http://www.microchem.com